> On Apr 3, 2024, at 1:49 PM, Rick Bensene via cctalk <[email protected]>
> wrote:
>
> ...
> Even with only having to render the digits zero through nine and a decimal
> point (the calculator didn't support negative numbers; they were represented
> using tens complement form), the display generator also used a batch of
> diode-transistor gates to generate the digits. The interesting thing about
> it is that instead of generating strokes to create the digits, the machine
> uses sine/cosine waveforms that are gated by the character generation logic
> to draw the digits on the screen. The position of the digits, like the CDC
> scopes, is derived by precision resistor DACs, and then a mixer takes over as
> the character is drawn using gated segments of the sine and cosine waveforms
> mixed together with the position voltage. The result is really beautifully
> rendered digits that look almost like they are drawn by a draftsperson who is
> extremely consistent in the drawing of each digit. The CRT has yellow-orange
> phosphor with a moderate persistence, so when the digits change, they look
> like they quickly morph from one digit to the next.
>
> The digits are among the nicest looking digits that I've ever seen on a CRT
> display, including those on the CDC scopes as well as IBM console displays.
I have, somewhere, a copy of a paper that describes analog circuits for
generating waveforms for digits along the lines you describe. Might have been
from MIT, in the 1950s, but right now I can't find it.
The CDC console waveforms start out as step function waveforms, with delta x
and/or y of +/- 1 or 2 units, at 100 ns intervals. Given the bandwidths of the
circuits involved they get rounded off in the generation, and a whole lot more
in the DD60 deflection amplifier signal chain. I've tried to create a SPICE
model of that signal path to try to reproduce what we know actually showed up
on the display screen, but haven't had much luck. Too much of the circuit
involves parts with unknown properties, starting with the transistors, on to
the wirewound resistors that apparently show up in various places, and ending
with the deflection plates of the CRTs themselves. Still, a crude IIR filter
mimicking some of the more obvious contributions do produce acceptable
character shapes in my DD60 emulation software.
Speaking of nice looking numeric displays: probably the best ever are the
projection displays made by IEE, in the 1960s I think.
https://www.antiqueradios.com/forums//viewtopic.php?f=12&t=341355 shows a
sample. A few computers from that era used them for the console, the CDC 1604
seems to be an example.
paul
